Background: Traditional ‘Framingham’ risk factors are widely used in prediction of coronary heart disease (CHD) and cardiovascular events both in patients with and without existing disease. Recent studies have confirmed poor prediction and lack of outcome change using such methods. The purpose of this study is to create a new model to stratify CHD risk using a broad selection of non-invasive and rapidly assessable variables. Methods: Seventy consecutive participants were recruited prior to elective coronary angiography for suspected CHD. Traditional risk factors, pulse wave analysis of the radial artery (PWA), 5-min heart rate variability (HRV), waist hip ratio and body-mass index (BMI) were measured in this analysis. Results: Good quality pulse waveforms were achievable within 2 weeks in 5 operators (mean quality index 89.7%±6.7%).Apart from a weak correlation between BMI and augmentation index of PWA, the four categories of risk variables were independent. Patients with triple-vessel disease (3VD) had significantly lower total power of HRV (mean 657ms2 versus 1485ms2 with <3VD, p < 0.01). Standard deviation of HRV and low frequency power were also significantly lower in logistic regression (p = 0.045, 0.025). Odds ratios for 3VD were 6.5 for pNN50 (p = 0.01) and 8.6 for total power (p = 0.03) at defined cutoffs. PWA and traditional risk variables did not reach statistical significance. Conclusion: Low 5-min HRV identifies patients with severe CHD. A multivariate model encompassing these variables plus cardiac biomarkers may improve prediction of the presence and severity of coronary disease. Clinical- Trials.gov Identifier NCT00403351.